The most important findings of the research presented here can be summarized as follows: (i) normal human keratinocytes display high levels of ERK activity that is sustained via the autocrine production of an EGFR ligand(s); (ii) keratinocytes respond to UVB irradiation with a stimulation of EGFR activity above the levels normally maintained by autocrine activation; (iii) the UVB-induced activation of EGFR is followed, unexpectedly, by a rapid and potent inactivation of ERK; (iv) the inactivation of ERK correlates with a UVB-induced (and yet-to-be-identified) signal transduction event that uncouples the EGFR-recruited ShcA and Grb2 from Ras; and (v) the UVB-induced inactivation of the Ras-ERK cascade in keratinocytes belongs to the more general ribotoxic stress response. To our knowledge, this is the first report showing that the functional activation of EGFR need not invariably result in an elevated ERK activity and that it may coincide even with an inhibition of ERK activity. We discuss here both the mechanisms of this inhibition and its potential physiological implications.
We have found that both HEKn and HEKn-E6/E7 appeared to maintain high basal activity of ERK even in the absence of exogenous growth factors (Fig. and ). The most likely reason for the high basal activity of ERK in proliferating keratinocytes is the autocrine production of soluble growth factor(s) (Fig. and ). Furthermore, based on the use of AG1478 and of the EGFR-neutralizing antibody, we concluded that these factors signal through EGFR. On the other hand, HaCaT cells, a human keratinocyte-derived cell line, (i) have lost the ability to produce an autocrine EGFR ligand(s) (Fig. and ), (ii) proliferate poorly even in complete keratinocyte medium (BKM+exoGF) (data not shown), and (iii) proliferate significantly better in the presence of growth factors supplied by fetal calf serum (data not shown). Taken together, these observations make a convincing argument that the autocrine production of an EGFR ligand(s) is critically important for the proliferation and homeostasis of human keratinocytes. Evidence from mice with manipulated egfr
alleles (hypomorphic, dominant-negative, or null) suggests that EGFR provides both proliferative and survival signals for the basal keratinocytes in mouse skin (reference 68
and references therein). It has been reported that keratinocytes produce four EGFR ligands, namely TGF-α (11
), amphiregulin (13
), HB-EGF (25
), and epiregulin (67
). We are currently investigating the identity of the factors that maintain the high basal levels of ERK activity in HEKn and HEKn-E6/E7.
Using NIH 3T3 mouse fibroblasts and HeLa cells, ERK was identified as the first MAP kinase whose activation was able to be detected early after irradiation with UVC radiation (55
). The identification of the ability of UVC to trigger the activation of EGFR (36
) gave further credence to the concept of an operational EGFR-ERK cascade as an important mediator of the UV-induced gene expression (6
). Evidence has been presented that the ability of UVC and UVB to activate EGFR resulted from the UV-induced incapacitation of protein phosphatases (36
). We demonstrate here for the first time that UVB causes the inactivation, rather than activation, of ERK in normal human keratinocytes. The wavelength of the radiation used (UVC versus UVB) cannot account for the differences in ERK behavior, since we have observed the same potent inactivation of ERK in keratinocytes with UVC and with UVB (data not shown) and since both UVC and UVB activate EGFR (36
). Therefore, we conclude that it is the cell type and not the type of UV radiation that determines the direction of the response of ERK.
A striking finding in our experiments was that the activation of EGFR in keratinocytes by UVB (Fig. ) did not result in the subsequent activation of Ras and ERK but rather in their inactivation. We propose that an important role in this process is played by a UVB-triggered event that uncouples the functionally EGFR-recruited adaptor proteins ShcA and Grb2 (Fig. ) from the activation of Ras (Fig. ). Grb2 contains one SH2 domain flanked by two SH3 domains (42
). The SH2 domain of Grb2 binds to tyrosine-phosphorylated residues in growth factor receptors, such as Y1068 and Y1086 of EGFR (5
), whereas the SH3 domains bind to proline-rich motifs found in many signaling molecules, such as Sos (9
). Sos is a guanine nucleotide exchange factor that greatly increases the activation of Ras in response to activation of growth factor receptors (9
). Since we failed to detect a UVB-induced disruption of the interaction of Sos with Grb2 and ShcA (Fig. ), it is likely that the Sos-Grb2 association is not the relevant target of the UVB-triggered inhibitory signal. It is possible that UVB interferes with the enzymatic activities of Sos and/or Ras GAP proteins (for a review, see reference 71
Since the UVB response of keratinocytes has both stimulatory (activation of SAPK) and inhibitory (inactivation of ERK) signal transduction components, it is reasonable to ask whether the two components originate from the same primary event (e.g., the interaction of UVB with the same sensory chromophore). Recent studies using nonepidermal cells have suggested that ribosomes actively engaged in translational elongation may be the cellular structures where the UV-induced signal leading to the activation of SAPK is initiated (31
). The results presented in Fig. and strongly suggest that both the stimulatory (activation of SAPK) and the inhibitory (inactivation of ERK) signal transduction components of the UVB response in keratinocytes require active ribosomes to occur. Interestingly, while the activation of SAPK by UVB is invariant in nonepidermal cells as well as in keratinocytes (31
), the UVB response of ERK is markedly different. In nonepidermal cells, UVC and UVB stimulated the activity of ERK in a way that required growth factor receptors but did not require active ribosomes (31
; unpublished data). In contrast, keratinocytes require EGFR to maintain the basal activity of ERK but require active ribosomes for the UVB-induced inhibition of ERK activity.
Although the UVB-induced inhibition of ERK is invariant in its occurrence in all experiments that we have performed in HEKn or HEKn-E6/E7, the duration of this inhibition was subject to some variation from experiment to experiment. Furthermore, while we have observed a precise temporal correlation between the onsets of inhibitions of Ras and of ERK whenever these analyses have been performed in parallel, we have observed that the basal activity of Ras tends to be restored faster than that of ERK. For instance, 2 h after irradiation of keratinocytes with 1,200 J/m2 of UVB, the activity of Ras was typically similar to the basal Ras activity (Fig. ). At the same time, the phosphorylation of ERK in most experiments did not reach basal levels until at least 4 to 6 h after the irradiation (Fig. ). These results suggest that the activity of ERK in keratinocytes may be regulated by UVB at a more complex level than the activity of Ras alone.
Observations with mice with a genetic inactivation of the egfr
gene have demonstrated the involvement of this receptor in the development of normal mouse skin (reference 68
and references therein). However, similar genetic evidence for the possible roles of ERK in the development of the mouse skin is still lacking. Particularly insufficient is our knowledge of the involvement of the EGFR-Ras-ERK cascade in the maintenance of skin homeostasis in long-lived mammals such as humans. However, indirect evidence strongly suggests that the proliferative state of basal keratinocytes in human skin crucially depends on the activity of the EGFR-Ras-ERK cascade. For instance, phosphorylated ERK has been detected, via immunohistochemical methods, almost exclusively in the stratum basale, the proliferative compartment of the skin harboring the keratinocyte stem cells (see below) (1
). Furthermore, abnormally increased phosphorylation of ERK in the stratum spinosum has been observed in cases of psoriasis, a disorder that is characterized by a keratinocyte hyperproliferation in the suprabasal layers (24
). Finally, patients enrolled in clinical trials of blocking anti-EGFR antibodies for the treatment of head-and-neck cancer displayed a marked decrease of ERK phosphorylation in the stratum basale. This decrease in ERK phosphorylation correlated with a dramatic decrease in the proliferation of basal keratinocytes (1
). In view of the abovementioned observations, we propose that the inactivation of the Ras-ERK pathway in proliferating basal keratinocytes (particularly in epidermal stem cells; see below) may provide an important initial signal arresting the cell cycle progression of cells that have not yet entered S phase (Fig. ). The model shown in Fig. is supported by the rapid (>50% within 2 h) decrease in the proportion of keratinocytes engaged in DNA synthesis after irradiation with UVB (unpublished data). It is thought that the major mediator of cell cycle arrest in UV-irradiated cells is the p53 tumor suppressor (29
). Surprisingly, we found that UVB irradiation of HEKn (p53-proficient cells [Fig. , lanes 1 to 5]) and HEKn-E6/E7 (p53-deficient cells [Fig. , lanes 6 to 10]) resulted in a similar rapid decrease in the [3
H]thymidine incorporation in both cell types (unpublished data). These results suggest that a p53-independent mechanism of cell cycle arrest is operational in UVB-irradiated keratinocytes. In support of the hypothesis that this p53-independent signal triggering a cell cycle arrest in irradiated keratinocytes may indeed be provided by the inhibition of ERK activity, we have found that the steady-state levels of cyclin D1 decline substantially in UVB-irradiated keratinocytes. It has been found that the expression of cyclin D1 requires the sustained activity of ERK (3
). Therefore, it is intriguing to speculate that even a transient inhibition of ERK activity in response to UVB may trigger a prolonged decline in the cyclin D1 levels (Fig. ).
FIG. 13. A model for the selective interference of UVB radiation with the EGFR-dependent signal transduction in keratinocytes. (A) In the absence of UVB, the autocrine stimulation of EGFR maintains an active Ras-ERK cascade that is involved in cell proliferation (more ...)
Epidermal stem cells (i.e., cells that remain in stratum basale after each cell division and retain their full proliferative potential) in the skin of long-lived mammalian organisms are likely to have developed unique ways to deal with repeated episodes of acute encounter with solar radiation. Suprabasal cells, such as differentiating or fully differentiated keratinocytes, can be eliminated by apoptosis (sunburn cells) without lasting consequences for the organism. However, shunting stratum basale stem cells into apoptosis after repeated episodes of acute irradiation may lead to a gradual depletion of the epidermis of stem cells, thereby severely hampering the continuous renewal of the skin. At the other extreme, tipping the balance toward increased survival after an encounter with a potent mutagen such as UVB carries the risk of a malignant transformation of the epidermal stem cell (thus leading to epidermal carcinomas). It seems likely that one sound strategy for fostering the survival of UVB-irradiated epidermal stem cells without the accumulation of mutations would be to arrest their cell cycle progression before they enter S phase (while actively suppressing apoptosis) until the DNA damage has been repaired. Upon completion of DNA repair, the epidermal stem cells (having been spared apoptotic cell death) can resume their important role of providing a continuous supply of differentiation-committed suprabasal keratinocytes. The newly described phenomenon of the UVB-induced uncoupling of the signaling pathway leading from EGFR to the activation of ERK presented here may suggest one possible mechanism executing exactly such a complex response to UVB. This hypothesis is outlined in Fig. . In proliferating basal keratinocytes, the low-level but persistent activation of EGFR (due to autocrine EGFR ligands) provides survival (antiapoptotic) and proliferative signals. The antiapoptotic program may be mediated through the sustained activity of the EGFR-PI3K-PKB/Akt1 cascade, whereas the mitogenic program may be mediated through the EGFR-Ras-ERK cascade (reference 68
and references therein) (Fig. ). By uncoupling EGFR from the Ras-ERK cascade, UVB may trigger the growth arrest program of the cell (Fig. ). At the same time, by stimulating the intrinsic tyrosine kinase activity of EGFR, UVB may potentiate the necessary survival signals, mediated by an EGFR-driven activation of PKB/Akt1, to prevent apoptosis (Fig. ). The experimental testing of this hypothesis is under way.
Finally, it should be noted that with regard to the responsiveness of ERK signaling to UVB, the HaCaT cell line behaves like fibroblasts and HeLa cells but unlike HEKn and HEKn-E6/E7. Cell lines selected by atypical culture conditions (e.g., serum-dependent growth in the case of HaCaT cells) thus may not provide useful (under some circumstances) models for the study of tissue-specific responses to toxic metabolites, carcinogens, and environmental stresses. The results presented here underscore the need for the development of suitable human tissue explants for studying these responses.